Metaheuristic vs Adaptive Approach in Discrete-Time Hammerstein Systems Identification

This paper aims to describe two identification methods for Hammerstein systems. Both methods are design to approximate the nonlinear component by using families of simpler functions. The first algorithm combines linear least squares with PSO to approximate both linear and nonlinear component parameters, whilst the latter redesigns the unknown coefficients approximation problem into a nonlinear least squares one and uses a modified version of Gauss-Newton algorithm to solve it. A comparison between the two methods is carried out.